As wireless computing devices become more ubiquitous, the demand for value-added mobile applications and solutions to deliver these applications increases. Enterprises expect new mobile applications to seamlessly integrate with the existing corporate information system, thus granting access to existing resources and services to highly mobile workers. Examples include accessing legacy applications and connecting to enterprise databases through wireless devices.
Mobile applications may have challenges such as:                Accessing a mobile application from different kinds of devices, for example from a PDA or a desktop computer, depending on the location of the user. PDAs and desktops differ in several ways, including form factors, operating systems and connectivity.        Running a mobile application on a mobile device continuing to work offline, in the case of a broken connection. An offline behavior utilizes access to local resources.        A mobile application accessing services published by some Web Services provider implementing Web Services standards, including SOAP, WSDL and UDDI.        
Deploying mobile applications in such heterogeneous environments greatly increases the complexity of the solutions, and presents many unresolved challenges. These challenges in turn affect the development of applications, as well as ongoing maintenance. Some issues to address are:                Open standards compatibility. Proprietary solutions are unlikely to be practical in such heterogeneous environments. The solution should fit the emerging Web Services standards, including SOAP, WSDL and UDDI.        Multi-platform deployment. The same service may be accessed from different kinds of devices and different communication protocols may be used.        Ease of modification. Since the markets change quickly, especially emerging markets like the wireless applications market, it is desirable that smart applications are easy and quick to modify.        Adaptable deployment. No single deployment model fits all contexts. The suitable deployment model depends on several factors, such as device resources, security requirements and application characteristics. Deployment models typically range from a thin client connected to a server, to a stand-alone running application.        Access to local resources. Mobile devices cannot rely exclusively upon server side resources, since the connection with a server cannot always be guaranteed. Furthermore, it can be costly to continually maintain a connection to a server when it is only occasionally required.        Enhanced user interface. Required to improve the user experience.        
Solutions like HyperText Markup Language (HTML) combined with HyperText Transfer Protocol (HTTP), Wireless Application Protocol (WAP), pure Java programming, Application Servers, and proprietary Software Development Kits (SDKs), each address a subset of these challenges, but more comprehensive solutions are desired.
HTML and WAP are open standards and may be deployed on multiple platforms. Services developed with these standards may be easily modified, since they become immediately available to the client devices when deployed on the server. Unfortunately, the deployment model of HTML and WAP is a rather rigid thin client model and a connection to a server must remain typically available during the execution of the application. This may be costly and even impractical, should the connection to the server be interrupted unexpectedly. Another drawback is that these solutions allow little access, if any, to local resources. Finally, the user interface is rather modest, at best.
Another solution mentioned above is pure Java programming. Although Java is quite portable, and is considered to some extent as an “open standard”, there are in fact multiple Java standards, including J2ME MIDP, Personal Java, and Java 2 Standard Edition. Thus, deploying a Java application on different platforms may require rewriting it several times, which implies involving highly-skilled developers. The same problem arises when modifying an application. Java provides good means of implementing a deployment model targeted to a specific architecture, accessing local resources and providing effective User Interfaces. Unfortunately, each deployment model generally requires a specific Java program, thus making it unfeasible to dynamically adapt a given application to new architectural requirements.
Proprietary SDK's provided by mobile device manufacturers are quite comparable in capabilities and drawbacks to the Java programming approach, but have the additional drawback of not being open standards. The use of such proprietary solutions entails a commitment to one particular (inflexible) solution, which in turn restricts the ability to later port an application to other mobile devices. Should the chosen solution at some point no longer meet the requirements of the user, it could be extremely costly to implement a completely new solution.
Application servers, and especially Java application servers based on the Java 2 Enterprise Edition (J2EE) standard, promote several architecturally significant separations of concerns. One architectural feature is a variation on the HTML approach, namely Java Server Pages (JSP). JSPs are targeted to improve the separation between interaction logic and business logic on the server side, but have the same limitations as traditional HTML architectures on the client side. Another architectural feature is the distinction between the development phase and the deployment phase for an application, thus allowing some flexibility for adapting to the underlying technical architecture. However, the deployment process is typically only concerned about server-side deployment characteristics, including transactions, security and database access. The distribution of processing and resources between the client and the server are not addressed by this solution.